KoreanAir Flight 0753

On December 26, 2018, KoreanAir flight 0753, an Airbus A220-300, registration HL8314, experienced a commanded inflight shutdown of the left (No. 1) engine, a Pratt & Whitney (P&W) PW1521G-3, while climbing through 29,000 feet on a flight from Busan, Korea (PUS) to Nagoya, Japan. The flight crew reported hearing a loud bang that was followed by vibrations and an engine fire warning. The flight crew shut down the engine and the airplane returned to PUS without further incident. The engine was removed from the airplane and shipped to P&W’s Columbus Engine Center, Columbus, Georgia for disassembly and examination.

The examination of the engine revealed one low pressure turbine (LPT) stage 3 blade, No. 52, that had a flat, planar, elliptical-shaped fracture surface at the rear half of the blade. The metallurgical examination of blade No. 52 showed that the grain size and hardness conformed to the requirements for the specified IN-100 nickel alloy. The energy dispersive spectroscopy (EDS) of blade No. 52 away from the origin of the fatigue crack produced a spectra that was consistent with the requirements for IN-100. However, the EDS of blade No. 52 at the preexisting intergranular crack and the origin of the fatigue crack produced a spectra that was consistent with IN-100, but also had peaks of zirconium and hafnium.

Arconic suggested the elevated level of zirconium that was noted in the EDS could be from the zirconia crucible in which the nickel alloy was melted. Since all of the blades from this engine and many others were cast from nickel alloy melted in a zirconia crucible and the Korean Air event was the only IN-100 LPT stage 3 blade fracture, that would suggest the elevated level of zirconium noted in the EDS of LPT stage 3 blade No. 52 was probably not a major factor in the failure of the blade. Hafnium is not one of the alloying elements in IN-100. It is however, one of the alloying elements of MAR-M-247, which according to Arconic’s records was the alloy that had been melted in the crucible just prior to it being used to melt the IN-100 nickel alloy to cast the blades that included the LPT stage 3 blade that would become No. 52. Arconic also suggested that the presence of the hafnium could be from the incomplete cleaning of the crucible from the previous melt that was of MAR-M-247 that has hafnium as an alloying element.

The presence of the hafnium in the intergranular crack would suggest the formation of a localized brittle structure that cracked very early in the life of the blade that progressed into the fatigue crack leading to the fracture of the blade and the subsequent loss of power. As previously noted, Arconic stated the crucible that was used to pour the IN-100 alloy into the molds for the LPT stage 3 blades including the blade that fractured had been used to pour MAR-M-247 nickel alloy. Although there are some common elements in both IN-100 and MAR-M-247, hafnium is unique to the MAR-M. Arconic asserted that the crucibles are cleaned between casting pours. However, the presence of the hafnium in the fractured blade would suggest that the cleaning of the crucible was inadequate.

According to Arconic, the blades undergo several post-casting inspections including visual, fluorescent penetrant inspection (FPI), and x-ray. The metallurgical examination of the fractured blade showed that the fatigue crack originated from a subsurface intergranular crack. Since the origin of the fatigue crack was a subsurface intergranular crack, presuming it was even cracked at the time of the post-casting visual inspections, it would not have been detectable since the visual and FPI inspections require a defect to be surface breaking to be detectable. Additionally, even if the defect was surface breaking, it is unlikely that the defect would be detectable. The initial intergranular crack probably only involved a few crystals that would be far smaller than 0.020-inch sized defect that would result in a 90 percent probability of detection. The blades also undergo an X-ray inspection. Although a subsequent review of the X-ray inspection records that was focused on the area where the fatigue crack originated detected a casting anomaly, the manufacturer stated that the casting anomaly lacked contrast to be detectable by the typical production level of inspection.

Verbatim NTSB analysis and probable cause from the NTSB dataset